GitLab 16.0 release notes

Unauthenticated users of the Projects List API will be subject to rate limitations moving forward.

On GitLab.com, the limit is set to 400 requests per 10 minutes per unique IP address.

Users of self-managed GitLab instances have the same rate limitation by default, but administrators can change the rate limits as they see fit. We encourage users who need to make more than 400 requests per 10 minutes to the Projects List API to sign up for a GitLab account.

Starting with 16.0, self-managed installations of GitLab will have two database connections by default, instead of one. This change makes self-managed versions of GitLab behave similarly to GitLab.com, and is a step towards enabling a separate database for CI features for self-managed versions of GitLab.

This change applies to installation methods with Omnibus GitLab, GitLab Helm chart, GitLab Operator, GitLab Docker images, and installation from source.

We have received feedback from users who wanted to prevent getting unwanted followers of their user profile. We listened to your concerns, so now, in your user profile settings under Preferences, you can disable following.

When you disable this feature, no one can follow you, and you cannot follow anyone. All existing following and follower relationships are removed, and the count is set to zero.

To prevent accidental deletion of projects and groups, starting in GitLab 16.0, the delayed deletion feature will be turned on by default for all GitLab Ultimate and Premium customers.

Self-managed users still have the option to define a deletion delay period of between 1 and 90 days, and SaaS users have a non-adjustable default retention period of 7 days.

Users of Ultimate and Premium groups can still delete a group or project immediately from the group or project settings via a two-step deletion process.

We believe that this change will contribute to a safer deletion process and will be beneficial in preventing accidental deletions. We’d love your feedback in issue #396996.

• Updates to GitLab 16.0 also update cert-manager to version 1.11.x. This cert-manager update includes breaking changes you must read before upgrading. These changes include a change to container names that was best done during a major release of GitLab. To see details of updated features, see the releases notes for cert-manager 1.11.
• PostgreSQL 12 is no longer supported. The minimum required version is PostgreSQL 13, and support for PostgreSQL 14 is added. New chart installs of GitLab include PostgreSQL 14 by default, and upgrades must follow the steps for upgrading the bundled PostgreSQL version.
• Updates to GitLab 16.0 include an update to the Redis subchart to version 16.13.2, including Redis 6.2.7.
• We have removed the bundled Grafana chart. For more information, see removal of the bundled Grafana Helm chart on our removals page. If you use the bundled Grafana, you must switch to the newer chart version from Grafana Labs or a Grafana Operator from a trusted provider.
• GitLab 16.0 includes registry services details for webservice and Sidekiq in the global.registry.* configuration for simplification because the values are present in both. You can keep the old behavior with an override.
• The minimum supported Helm version is 3.5.2.
• The GitLab Runner default version is now Ubuntu 22.04.

• PostgreSQL 12 is no longer supported. The minimum required version is PostgreSQL 13. Users of the packaged PostgreSQL 12 must perform a database upgrade before installing GitLab 16.0.
• The new base OS for the Omnibus GitLab docker images is Ubuntu 22.04.
• GitLab 16.0 disables older telemetry endpoints for Consul, which were deprecated in Consul 1.9. This allows us to update Consul to newer versions.
• GitLab 16.0 includes packages for Red Hat Enterprise Linux (RHEL) 9 and compatible distributions.
• GitLab 16.0 includes Mattermost 7.10 with security updates. An upgrade from earlier versions is recommended.

GitLab Free customers with a self-managed instance running GitLab Enterprise Edition can now access five more paid features under the Registration Features program:

• Password complexity policy
• Description change history
• Issue board configuration
• Maintenance mode
• Coverage-guided fuzz testing

To get access to these features, register with GitLab and send us activity data through Service Ping.

In GitLab 15.10, we started mapping GitHub repository collaborators as GitLab project members during GitHub project imports. We received feedback that this led to confusion and that some GitHub collaborators were unexpectedly added and consumed seats.

In GitLab 16.0, we’ve iterated and added GitHub repository collaborators to the list of additional items to import. This gives users the option to avoid importing these users and to understand the possible implications of importing them.

This option is selected by default. Leaving it selected might result in new users using a seat in the group or namespace, and being granted permissions as high as project owner. Only direct collaborators are imported. Outside collaborators are never imported.

If you own or collaborate on a lot of repositories in GitHub, you might have trouble finding those that you want to import to GitLab using the current filtering option.

To make finding the right repositories easier, we have added additional filters. You can now list subsets of the repositories you can import using three tabs:

• Owner, to list repositories you own.
• Collaborator, to list repositories you collaborate on.
• GitHub organization, to list repositories that belong to GitHub organizations.

On the Organization tab, you can further narrow down your search and choose a specific organization and list only repositories belonging to that organization.

When a user raises an access request for a group or project, the request appears in the To-Do List of the group or project owner. For groups and projects that have multiple owners, the request appears in each owner’s To-Do List.

With this new functionality, to-do items that have already been completed by another owner are marked Done in the others’ To-Do Lists.

GitLab 16.0 features an all-new navigation experience! To get started, go to your avatar in the top right of the UI and turn on the New navigation toggle. The left sidebar changes to a new and improved design, based on user feedback we’ve received over the last year.

Please let us know about your experience in this issue. Based on the feedback, we will be progressively enabling the new navigation across our user base, with the final step being removal of the old navigation.

Previously, you could only authenticate the Jira issue integration with a Jira username and password.

Now you can use a Jira personal access token to authenticate if you are using Jira Data Center and Jira Server with Jira 8.14 and later. A Jira personal access token is a safer alternative to a username and password.

It is useful for a Service Desk requester to see their original request in the automated thank you email replies.

In this release, we add an %{ISSUE_DESCRIPTION} placeholder so that Service Desk administrators can include the original request in the thank you email.

When working on merge requests, it’s important to make sure that what you’re seeing is the latest information for approvals, pipelines or other information that might impact your ability to get the changes merged. Historically, this has meant refreshing the merge request or waiting for polling updates to come through.

We’ve improved the experience of both the merge button widget and approval widget inside of the merge request, so that they now update in real-time in the merge request. This is a great improvement to improve the speed at which you can deliver changes, and the confidence at which you can move a merge request forward knowing you’re seeing the latest information.

We’re looking at more areas for real-time improvements in merge requests, so follow along for updates.

When selecting one or more vulnerabilities in the vulnerability report, it’s possible to change their status in bulk.

With this release, you can now select a dismissal reason when choosing the dismiss status, and add a comment when changing a vulnerability’s status."

In GitLab 15.11, we added bulk adding and removing of compliance frameworks to the compliance frameworks report.

Now in GitLab 16.0, you can also add and remove compliance frameworks from projects directly from the report table row.

Before GitLab 16.0, you had to create and edit frameworks in the group’s settings.

Now in GitLab 16.0, you can create or edit your compliance frameworks in the compliance framework report as well. This simplifies the framework creation workflow and reduces the need to switch contexts while managing your frameworks.

Prior to GitLab 16.0, the compliance violations report showed all violations on all branches.

Now you can now filter violations using the new Search target branch field, allowing you to focus on the branches that you are most concerned with.

With role-based approval actions, you can configure scan result policies to require approval from GitLab-supported roles, including Owners, Maintainers, and Developers.

This gives you additional flexibility over requiring individual approvers or defined groups of users, making it easier to enforce policies based on roles you already leverage in GitLab, at scale, especially across large organizations.

Previously, GitLab’s DAST analyzers did not support callback attacks while performing active checks. This meant that Out-of-band Application Security Testing (OAST) needed to be configured separately from your DAST scan.

Now, you can run OAST by extending the browser-based DAST analyzer configuration to enable callback attacks.

In this release we are introducing the BAS.latest.gitlab-ci.yml template. The Breach and Attack Simulation CI/CD template features job configuration for the browser-based DAST analyzer and enables container-to-container networking to add extended DAST scans against service containers to your CI/CD pipeline.

We’re continuously iterating to develop new Breach and Attack Simulation features. We’d love to hear your feedback on the addition of callback attacks to browser-based DAST.

Have you been thinking about moving your Maven or Gradle repository to GitLab, but haven’t been able to invest the time to plan the migration? GitLab is proud to announce the MVC launch of a Maven/Gradle package importer.

You can now use the Packages Importer tool to import packages from any Maven/Gradle compliant registry, like Artifactory.

To use the tool, simply create a config.yml file that contains the details of the packages you want to import into GitLab. Then add the importer to a .gitlab-ci.yml pipeline configuration file, and the importer does the rest. It runs in the pipeline, dynamically generating a child pipeline with jobs that import all the packages into your GitLab package registry.

The GitLab Package Registry now supports downloading Maven packages using the Scala build tool (sbt). Previously, Scala users had no way to download Maven packages from the registry because basic authentication was not supported. As a result, Scala users were either blocked from using the registry or had to use Maven (mvn) or Gradle as an alternative.

By adding support for Scala, we hope to help you use the Package Registry with your more data intensive projects.

Please note that publishing artifacts using sbt is not yet supported, but you can follow issue 408479 if you are interested in adding support for publishing.

We know that GitLab To-Do List is a widely adopted feature, but it was not available on tasks, objectives, and key results.

In this release, we’re introducing the ability to toggle a to-do item on or off from a work item record.

In previous versions of GitLab, cookies of different GitLab Pages sites under the same top-level group were visible for other projects under the same top-level because of the GitLab Pages default URL format.

Now, you can secure your sites by assigning a unique subdomain to each GitLab Pages project.

You can now contribute to tasks, objectives and key results with the addition of emoji reactions for work items.

Before this release, you could only add reactions on issues, merge requests, snippets, and epics.

Until now, you could specify only a fixed number of colors for your labels.

This release introduces a color picker to label management, allowing you to select any range of colors for your labels.

If you’re a user of tasks or OKRs you’ve likely wished more than once that we could reorder the child records within the widget!

With this work, users will now be able to reorder child records within work item widgets allowing them to indicate relative priority or signal what’s up next.

Value Stream Analytics has been extended with two new stage events: issue first assigned and merge request first assigned. These events can be useful for measuring the time it takes for an item to be first assigned to a user.

To implement this feature, GitLab started storing the history of assignment events in GitLab 16.0. This means that issue and MR assignment events prior to GitLab 16.0 are not available.

GitLab SAST includes many security analyzers that the GitLab Static Analysis team actively maintains, updates, and supports. We published the following updates during the 16.0 release milestone:

• The Semgrep-based analyzer includes updated GitLab-managed scanning rules. See the CHANGELOG for further details. We’ve updated the rules to:
  • Update OWASP mappings to show that they’re based on the 2017 OWASP Top Ten. Thanks to @artem-fedorov for this community contribution.
  • Handle additional cases in the PyYAML.load rule. Thanks to @stevep-arm for this community contribution.
  • Significantly improve the descriptions and guidance for C rules based on revisions from the GitLab Vulnerability Research team.
  • Add support for scanning Scala code.
• The Flawfinder-based analyzer now supports passing the --neverignore flag to disregard “ignore” directives in comments. See the CHANGELOG for further details.
• The KICS-based analyzer is updated to KICS version 1.7.0. See the CHANGELOG for further details.
• The MobSF-based analyzer now supports multiple modules and projects, which resolves several bug reports. See the CHANGELOG for further details.

Also, as previously announced, we increased the major version number of each analyzer as part of GitLab 16.0.

If you include the GitLab-managed SAST template (SAST.gitlab-ci.yml) and run GitLab 16.0 or higher, you automatically receive these updates. To remain on a specific version of any analyzer and prevent automatic updates, you can pin its version.

For previous changes, see last month’s updates.

We regularly release updates to the GitLab Secret Detection analyzer. During the GitLab 16.0 milestone, we:

• Added GitLab-managed detection rules for:
  • Access tokens for the Meta, Oculus, and Instagram APIs.
  • Tokens for the Segment Public API.
• Updated the Gitleaks scanning engine to version 8.16.3.
• Fixed a bug that prevented scanning when a repository had only a single commit.
• Incremented the analyzer major version to 5, as previously announced.

See the CHANGELOG for further details.

If you use the GitLab-managed Secret Detection template (Secret-Detection.gitlab-ci.yml) and run GitLab 16.0 or higher, you automatically receive these updates. To remain on a specific version of any analyzer and prevent automatic updates, you can pin its version.

For previous changes, see last month’s updates.

We have optimized the way that the browser-based DAST analyzer performs its scans. These improvements have significantly decreased the amount of time that it takes to run a DAST scan with the browser-based analyzer. The following improvements have been made:

• Added log summary statistics to help determine where time is spent during a scan. This can be enabled by including the environment variable DAST_BROWSER_LOG="stat:debug".
• Optimized passive checks by running them in parallel.
• Optimized passive checks by caching regular expressions used when matching content in HTTP response bodies.
• Optimized how DAST determines whether a page has finished loading. Now, we don’t wait for excluded document types or out-of-scope URLs.
• Reduced waiting time for pages where the DOM stabilizes quickly after page load.

With these improvements, we have seen browser-based DAST scan times reduced by 50%-80%, depending on the complexity and size of the application being scanned. While this percentage decrease may not be seen in all scans, your browser-based DAST scans should now take significantly less time to complete.

GitLab Static Application Security Testing (SAST) now offers Semgrep-based scanning for Scala code. This work builds on our previous introduction of Semgrep-based Java scanning in GitLab 14.10. As with the other languages we have transitioned to Semgrep-based scanning, Scala scanning coverage uses GitLab-managed detection rules to detect a variety of security issues.

The new Semgrep-based scanning runs significantly faster than the existing analyzer based on SpotBugs. It also doesn’t need to compile your code before scanning, so it’s simpler to use.

GitLab’s Static Analysis and Vulnerability Research teams worked together to translate rules to the Semgrep format, preserving most existing rules. We also updated, refined, and tested the rules as we converted them.

If you use the GitLab-managed SAST template (SAST.gitlab-ci.yml), both Semgrep-based and SpotBugs-based analyzers now run whenever Scala code is found. In GitLab Ultimate, the Security Dashboard combines findings from the two analyzers, so you won’t see duplicate vulnerability reports.

In a future release, we’ll change the GitLab-managed SAST template (SAST.gitlab-ci.yml) to only run the Semgrep-based analyzer for Scala code. The SpotBugs-based analyzer will still scan code for other languages, including Groovy and Kotlin. You can disable SpotBugs early if you prefer to use only Semgrep-based scanning.

If you have any questions, feedback, or issues with the new Semgrep-based Scala scanning, please file an issue, we’ll be glad to help.

Previously, a trigger job configured with strategy: depends mirrored the job status of the downstream pipeline. If the downstream pipeline was in the running status, the trigger job was also marked as running. Unfortunately, if the downstream job did not comnplete and had a status canceled, the trigger job’s status was inaccurately failed.

In this release, we have updated trigger jobs with strategy: depend to reflect the downstream’s pipelines’s statis accurately. When a downstream pipeline is cancelled, the trigger also shows canceled.

This change may have an impact on your existing pipelines, especially if you have jobs that rely on the trigger job’s status being marked as failed. We recommend reviewing your pipeline configurations and making any necessary adjustments to accommodate this change in behavior.

In this release we are excited to announce the availability of CI/CD components, as an experimental feature. A CI/CD component is a reusable single-purpose building block that can be used to compose a part of a project’s CI/CD configuration, or even an entire pipeline.

When combined with the inputs keyword, a CI/CD component can be made much more flexible. You can configure the component to your exact needs by inputting values which can be used for job names, variables, credentials, and so on.

Using a cache is a great way to speed up your pipelines by reusing dependencies that were already fetched in a previous job or pipeline. But when there is no cache yet, the benefits of caching are lost because the job has to start from scratch, fetching every dependency.

We previously introduced a single fallback cache to use when no cache is found, that you can define globally. This was useful for projects that used a similar cache for all jobs. Now in 16.0 we’ve improved that feature with per-cache fallback keys. You can define up to 5 fallback keys for every job’s cache, greatly reducing the risk that a job runs without a useful cache. If you have a wide variety of caches, you can now use an appropriate fallback cache as needed.

The include keyword lets you compose your CI/CD configuration from multiple files. For example, you can split one long .gitlab-ci.yml file into multiple files to increase readability, or reuse one CI/CD configuration file in multiple projects.

Previously, a single CI/CD configuration could include up to 150 files, but in GitLab 16.0 administrators can modify this limit to a different value in the instance settings.

In this new workflow, adding a new runner to a project requires authorized users to create a runner in the GitLab UI and include essential configuration metadata.

With this method, the runner is now easily traceable to the user, which will help administrators troubleshoot build issues or respond to security incidents.

Previously, the GET /api/:version/projects/:id/jobs was rate limited to 2000 authenticated requests per minute.

To move this in line with other rate limits and improve efficiency and reliability, we have lowered the limit to 600 authenticated requests per minute.

We’re also releasing GitLab Runner 16.0 today! GitLab Runner is the lightweight, highly-scalable agent that runs your CI/CD jobs and sends the results back to a GitLab instance. GitLab Runner works in conjunction with GitLab CI/CD, the open-source continuous integration service included with GitLab.

What’s new:

• GitLab Runner autoscaling plugin for Google Compute Engine - Experiment

The list of all changes is in the GitLab Runner CHANGELOG